#!/usr/bin/Rscript

# Figure: QUAST statistics

############################## LOG
sink(file=file(snakemake@log[[1]], open="wt"), type=c("output", "message"))

############################## LIBS
suppressMessages(library(testit)) # assertions
suppressMessages(library(dplyr)) # tables
suppressMessages(library(patchwork)) # combine plots

# custom
source(snakemake@params$const)
source(snakemake@params$utils)

############################## CONFIG

# assemblers
ASM_NAMES_CONFIG <- c(
    snakemake@config$assemblers$lr,
    snakemake@config$assemblers$sr,
    snakemake@config$assemblers$hy
)
ASM_TOOL_NAMES  <- ASM_TOOL_NAMES[ASM_NAMES_CONFIG]
ASM_TOOL_COLORS <- ASM_TOOL_COLORS[ASM_NAMES_CONFIG]; names(ASM_TOOL_COLORS) <- ASM_TOOL_NAMES[names(ASM_TOOL_COLORS)]
ASM_TOOL_SHAPES <- ASM_TOOL_SHAPES[ASM_NAMES_CONFIG]; names(ASM_TOOL_SHAPES) <- ASM_TOOL_NAMES[names(ASM_TOOL_SHAPES)]

############################## DATA

# individual tables
TABS <- list()
for(sname in names(snakemake@config$samples)){
    testit::assert(sname %in% names(snakemake@input))
    TABS[[sname]] <- read_prodigal_gcounts(snakemake@input[[sname]])
}

############################## STATS

print("STATS: Number of total proteins: max vs min:")
for(sname in names(snakemake@config$samples)){
    print(sprintf(
        "STATS: Sample %s: %s vs %s = %d / %d = %.2f",
        sname,
        TABS[[sname]]$tool[which.max(TABS[[sname]]$total)],
        TABS[[sname]]$tool[which.min(TABS[[sname]]$total)],
        max(TABS[[sname]]$total),
        min(TABS[[sname]]$total),
        max(TABS[[sname]]$total) / min(TABS[[sname]]$total)
    ))
}

############################## PLOT

# individual plots
FIGS <- list()
for(sname in names(snakemake@config$samples)){
    FIGS[[sname]] <- plot_prodigal2(TABS[[sname]]) +
    labs(
        title=LETTERS[which(names(TABS) == sname)],
        subtitle=SAMPLE_NAMES[sname]
    )
}

# combine plots
FIGS_ALL <- Reduce("+", FIGS) + plot_layout(ncol=2)

# PDF
pdf(snakemake@output$pdf, width=20, height=14)
print(FIGS_ALL)
dev.off()